Improve execution time for this flow

[Anonymous]

Hello everyone!

I have seen many topics about how to improve your execution time, but none of them work with my flow.

My flow gets the text from resumes which are PDF documents, and other information such as keywords from Excel files. The goal is to fill another Excel file with chosen information from the PDF documents. It has to save the name of every company the person has worked for, for how long, and look for the keywords in key places. It has to do date conversions and calculate their experience, in total and in certain fields based on the keywords. If it's not clear enough feel free to ask ! 

I have looked through topics about the improvement of execution time, and the only ones I have found useful said to reduce the run time between actions, and none of them have an option that can be unchecked to go faster (like some tips I have seen said).

I have to use this flow for 1000 documents and with 237 keywords. I have only done performance tests for 5 documents and 237 keywords, which took more than 3 hours. It would then take 25 days to be complete with 1000 documents and 237 keywords. I have also done a similar flow on UiPath, another way to do RPA, and it takes 2 hours for 1000 documents and 237 keywords.

So I would like to know if there is any way to improve the execution time for this type of flow? As PAD is a great RPA developer, I would love to first find a solution before settling in using UiPath.

Thank you very much in advance !

[amamidwar23]

Hi @Anonymous 

yes execution time for actions in PAD is more than UiPath.

hope you have checked which part of actions are taking more time in PAD?

Assuming it is excel  feeding activity , then you can try VBA , which will more faster and quick.

 I have had created a vba where  reading  , searching, matching keywords from a 100 PDF pages and feeding them to excel took 5 minutes 

Could you elaborate what exactly the keywords from excel playing part of .

Thanks 

AR

[Anonymous]

Hello! 

thank you for your reply.

The part that takes more time is looking up if a keyword is found or not because there are a lot of keywords, so the flow has to go through the loop a lot of times. 

I'm sorry but what is VBA ?

On Excel, a number will be added whether a keyword is found or not: 1 if it is found, 2 if not. A lot of the keywords are software names, like Java, Php ...

Thank you

Noémie

[amamidwar23]

HI @Anonymous 

VBA stands for visual basic for applications , uses vb scripts for automating tasks over excel .

reading PDF for keywords can be achieved using VBA through ADOBE acrobat COM API's. quick and fast, that is alogether different approach , which i can help you with, if you want.

For PAD, to understand your flow, could you help me with snip of actions where you are trying to read PDF files ?

i'll look into it .

Regards,

AR

[Anonymous]

Thank you @amamidwar23 !

Okay, I don't quite understand what VBA is but I will look into it, thanks!
But do you think it could be done using WORD documents and not PDF? The documents we use are Word, but we couldn't find a way to retrieve the exact text from word documents (do you know if it is possible by going premium ?). We settled by converting them to PDF using Power Automate Cloud because the text could easily be retrieved from this type of documents with PAD.

Yes! Here is the snip of my actions on how we first read the text from the pdf and then how we use it to check for keywords: : (I'm sorry it's in French but I can help if you need help understanding) :

[Daniel_Pa]

Did you figure this out?

How did you run your "test" to determine the speed? I've noticed that having the process designer open, will greatly decrease the speed of excel-actions. Try running the the flow from the console, to determine the actual speed.

Furthermore, beware of having a delay between actions when testing through the process designer. At the bottom, you can set a static delay, which will be used between each action.

The default is 100ms, which will greatly increase the processing time with large/many loops.

[Anonymous]

Hello @Daniel_Pa ! 

No we still haven't figured this out ! So thank you for your answer ! 

To get the time between the parts, I have put action to get the current time and then subtract those times to get the execution time on different parts of the robot. Yes i always ran the test with the process designer open, I will try with the console close !

Yes I have noticed this with the run delay, so I have tried putting it to 1ms. It in fact decreases the time, but not a lot or at least no enough...

[Daniel_Pa]

To avoid increasing time with extra steps (writing time codes to files), try instead to run the flow on a limited number of files.

Perhaps you could create a subfolder with just 20 files and run the flow on that folder, instead of all the files - and to be able to do it without the designer open.

And of course omit the actions you use to log the time and instead just track when you start it and when it ends, an then calculate the run time per file.

That should give you a better idea of the actual run time.

[yoko2020]

1. Don't run the process with debugger, run it without debugger instead.

2. Use high end machine with 8 core and 16 GB RAM, and feel the difference.

[tkuehara]

Hi @Anonymous 

I've compared WinAutomation and Power Automate Desktop and any Excel-related action takes double the time in PAD. So this is by design. If you want to optimize that timing, I suggest changing the approach to an "Excel as Database" one. Basically you'll read, insert and/or modify values in an Excel file through a SQL command. This is by far the best way I found to improve Excel performance issues in PAD.

I'm using an old version (2010), but the connector should look something like this:

Under "Connection" tab, you point to your Excel file:

Note: you'll need the "Microsoft Access Database Engine" installed in your machine to make it work.

As for reading values from Word directly, that's not possible. The UI Selector is unable to go down beyond the "Edit" field: